Alcohol Doesn't Ruin Your Sleep The Way You Think

A piece of data that we sometimes have access to includes knowing when a user had alcohol before going to bed. Last Wednesday, I sat down and cleaned this data to see what I could find. The result? 15,000 nights of sleep. Somehow, we’ve found ourselves in a unique position that allows us to investigate the effects of drinking before sleep quite rigorously.

We ran this data through a mixed effects model, and analyzed the statistically significant effects. Broadly, drinking before you sleep wrecks your sleep. You often spend more time awake in bed, your recovery is worse, you wake up more often, and your breathing becomes weird. But you probably guessed this. So let’s have a look at the numbers.

For time spent awake in bed, we saw an effect of 398 seconds, with a standard error 116 seconds for extra time spent awake in bed. When I first saw this, I was satisfied. It was an effect that made sense. To have a deeper look, I plotted the distribution of these awake times. Turns out, I was losing quite a bit of nuance.

alc_awake_dist (1).png — **Figure 1:** Distribution of time spent awake in bed.

Through our mixed effects model, we know that alcohol does overall make you spend more time awake in bed. However, the distribution does tell us something more. The spike at ~1000 seconds tells us that alcohol widens the spread of time taken to fall asleep. So while most people would fall asleep within the first 15 minutes without alcohol, this guarantee vanishes when alcohol is consumed before going to bed.

The only guarantee that remains is that you will almost definitely spend more time awake. Figure 2 shows this more cleanly, comparing means between users that logged both non-alcohol and alcohol nights.

**Figure 2:** Within user means of time spent awake in bed on alcohol and non-alcohol nights.

Let’s move and have a look at the quality of sleep. Through our model we saw an increased number of wakeup events on nights where alcohol was drank. When looking at proportions of number of wakeup events within the alcohol and no-alcohol groups, this effect of alcohol disrupting sleep remained consistent.

**Figure 3:** Distribution of number of wakeup events on alcohol vs non-alcohol nights.

Surprisingly, we saw no significant effects in the stages of sleep: light, deep, and REM. This was an effect I had expected to see, particularly due to the changes in breathing we saw in our analysis. There’s two main values we look at when assessing breathing during sleep; blood oxygenation and respiratory rate. We excluded a reduced blood oxygenation effect from this report due to a p-value of 0.06.

However, respiratory rate shows a much clearer link, with a p-value of 0.02. Specifically, we saw an increase of 0.17, with a standard error 0.07. This shows up quite clearly in within-user differences, as shown below.

**Figure 4:** Within user means of respiratory rate while asleep on nights with alcohol vs no alcohol.

While increased respiratory rates are not strictly a bad thing, it indicates that your body is trying harder to get oxygen and get rid of carbon-dioxide. If alcohol pushes this increase past 20 breaths per minute, it becomes a big cause of concern!

Putting these findings together, we come to the conclusion that if you consume alcohol at night, it make you sleep badly in almost every way. You have a hard time falling asleep. you wake up more and your body has to try harder to meet its oxygen demand.

Bonus results: We found that you nap less on days where you decide to drink alcohol before going to bed! Very strange and perhaps a case of statistics hiding more than they reveal. We also saw increased step counts on days where users recorded alcohol before bed—maybe a walk to pub!